Real-Time Yoga Pose Detection Using OpenCV and MediaPipe
Isha Chaudhary, Nongmeikapam Thoiba Singh, Mahak Chaudhary, Komal Yadav
Abstract
In this era of advanced technology, where everything is going online, from shopping to education, there may be ways to get an automated, personalised trainer and mentor for yoga and meditation. Yoga is one of the best exercises for maintaining good physical and mental health. But what if yoga is done incorrectly? It can be harmful to our bodies, so there is a need for a personalised yoga trainer and mentor who is available 24 x 7 to assist you with the poses and provide you with immediate feedback in order to maintain a healthy and wealthy body. In this model, we are coming up with the idea of providing real-time posture detection using deep learning and computer vision that will guide the user to the correct postures after comparing them with the standard yoga postures and also help them track their gym activities. Our research also aims to integrate all of the features from various platforms related to posture detection applications on a single platform rather than switching between multiple platforms. In order to make it more user-friendly, we are giving users the option of selecting between an AI tutor and a physical trainer based on their needs. One of the objectives of the model is to display information about the pose, such as the benefits of performing it and how to incorporate it into routine exercises that the trainee can simultaneously practise to increase its effectiveness.